基于小波变换的X射线焊缝图像缺陷识别方法研究  

Research on X-ray weld image defect recognition method based on wavelet transform

作　　者：李波 樊磊 高正浩 刘映霞[2] 潘国熙 张辉 LI Bo;FAN lei;GAO Zhenghao;LIU Yingxia;PAN Guoxia;ZHANG Hui(Electric Power Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 550002,China;Guizhou Polytechnic College of Communications,Guiyang 55000,China;Liupanshui Power Supply Bureau,Guizhou Power Grid Co.,Ltd.,Anshun,Guizhou 561000,China;Anshun Power Supply Bureau,Guizhou Power Grid Co.,Ltd.,Anshun,Guizhou 562400,China)

机构地区：[1]贵州电网有限责任公司电力科学研究院,贵阳550002 [2]贵州交通职业技术学院,贵阳550002 [3]贵州电网有限责任公司六盘水供电局,贵州安顺561000 [4]贵州电网有限责任公司安顺供电局,贵州安顺561000

出　　处：《自动化与仪器仪表》2020年第3期63-67,共5页Automation & Instrumentation

基　　金：南方电网网级决策项目“GIS设备数字射线智能诊断与专家会诊系统关键技术研究”资助。

摘　　要：经过长时间的研究发现,在焊缝图像缺陷识别中,传统的方法具有正确识别率低的问题,为此提出了基于小波变换的图像缺陷识别方法。输入获取的X射线焊缝初始图像信息,从帧数叠加、数字形态学变换和图像增强三个方面对初始图像进行预处理,约束小波变换阈值降低图像噪声,最后通过边缘检测,提取焊缝图像当中的缺陷特征。在对比实验当中,设立两种传统识别方法作为实验的对照组,提出的X射线焊缝图像缺陷识别方法为实验组,同时对统一型号的焊接工件进行缺陷识别,实验发现提出的识别方法的正确识别率高达91.8%。After a long time of research,it is found that the traditional method has the problem of low correct recognition rate in weld image defect recognition.Therefore,an image defect recognition method based on wavelet transform is proposed.Inputing the acquired initial image information of X-ray welding seam,the initial image is preprocessed from three aspects:frame number superposition,digital morphological transformation and image enhancement,the wavelet transform threshold to is restrained reduce image noise,and finally the defect features are extraeted of the welding seam image through edge detection.In the comparative experiment,two traditional recognition methods are set up as the control group.The proposed X-ray weld image defect recognition method is the experimental group.At the same time,defect recognition of uniform welded workpiece is carried out.The experimental results show that the correct recognition rate of the proposed recognition method is up to 91.8%.

关 键 词：小波变换 X射线 图像识别 识别方法 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]